Appendix¶
Video frames for model training¶
10% of the training set is used as evaluation set
4K:
Bund_Nightscape frame 0,
Campfire_Party frame 0, 10
Construction_Field frame 0, 250
ducks_take_off_2160p50 frame 0
Fountains frame 0, 12
Library frames 0, 103
Marathon frame 1
old_town_cross_2160p50 frame 3
park_joy_2160p50 frame 3
Residential_Building frame 0, 120
Runners frame 0, 7
Rush_Hour frame 0, 10
Scarf frame 0
Tall_Buildings frame 0
Traffic_and_Building frame 0, 145
Traffic_Flow frame 0, 10
Tree_Shade frame 0
Wood frame 0, 81
1080p:
crowd_run_1080p50 frame 0, 40
dinner_1080p30 frame 241, 750
factory_1080p30 frame 711
old_town_cross_1080p50 frame 3
riverbed_1080p25 frame 3
rush_hour_1080p25 frame 69
tractor_1080p25 frame 480, 662
720p:
FourPeople_1280x720_60 frame 0
Johnny_1280x720_60 frame 0
old_town_cross_420_720p50 frame 3
vidyo4_720p_60fps frame 0
Total number of samples:
64x64: 63,180
32x32: 252,336
16x16: 978,745
Video frames for test¶
1080p:
BasketballDrive_1920x1080_50 frame 0
BQTerrace_1920x1080_60 frame 0
Kimono1_1920x1080_24 frame 1
ParkScene_1920x1080_24 frame 1
station2_1080p25 frame 1
sunflower_1080p25 frame 1
Tennis_1920x1080_24 frame 1
720p:
720p50_mobcal_ter frame 500
720p50_parkrun_ter frame 2
720p50_shields_ter frame 0
720p5994_stockholm_ter frame 2
KristenAndSara_1280x720_60 frame 0
park_joy_420_720p50 frame 499
vidyo3_720p_60fps frame 0
vidyo4_720p_60fps frame 29
small training set:
Bund_Nightscape frame 0,
Campfire_Party frame 0
Construction_Field frame 0
Fountains frame 0
old_town_cross_2160p50 frame 3
Residential_Building frame 0
Runners_frame 0
Traffic_and_Building frame 0
Traffic_Flow_frame 0
Tree_Shade_frame 0
Wood_frame 0
